We investigated the roles of particulate matter with unstable implant, in engendering the aggressive tissue response associated with implant loosening in humans. This study serves as a basis for establishing a controlled animal model to reproduce the conditions present after implant loosening. The model includes a 6 mm polymethylmethacrylate (PMMA) cylinder concentrically pistoning 500 microm under load in a 0.75-mm circumferential gap, inserted into canine medial femoral condyles for 8 weeks. We evaluated two size concentrations of polyethylene: type A particulate polyethylene (0.5-12 microm), and type B particulate polyethylene (0.5-50 microm; 85% < 12 microm). The following three treatment groups were investigated in 28 unstable implants in 14 dogs: (1) without polyethylene (control), (2) with type A polyethylene, and (3) with type B polyethylene. We found an aggressive periprosthetic membrane, similar to that seen at revision in humans, only in the unstable implant with polyethylene. The features of this membrane included macrophages with intracellular polyethylene, a dense fibrous membrane with a synovial-like lining layer, and a sclerotic neocortex. The size distribution of the polyethylene did not alter the tissue response. An unstable implant without polyethylene resulted in a benign, quiescent membrane with loose fibrous connective tissue. The model creates a revision cavity analogous to that seen in revision joint arthroplasty, and merits further studies of revision joint replacement.